摘要
针对气象变化时负荷曲线预测精度低、预测模型不能完全适应气象变化的情况,提出了一种基于模糊信息粒化与多策略灵敏度的短期日负荷曲线预测方法。提出了完全气象因子序列的概念,建立气象粒化集;采用空间多元回归及滞后模型结合多策略灵敏度分析法,建立了针对复杂气象条件下的极值预测模型;基于改进的K-means聚类分析法查找并获取气象特征日,计算初步预测曲线,主动判断预测曲线畸变概率并进行优化修正,得到最佳预测日负荷曲线;利用动态数据流对模型参数进行更新,实现精细化预测。最后采用该方法对我国南方某地区全年负荷曲线进行预测,验证了模型在多种气象条件下的预测准确性,尤其适用于短期内气象存在复杂变化的情形。
Due to the variable weather,the precision of short-term load forecasting curves and forecasting model can not fit the demand of power grid. This paper proposes a short-term daily load curve forecasting method based on fuzzy information granulation and multi-strategy sensitivity. The concept of global meteorology factor is introduced,and used to build up the meteorological granulating set. Spatial multiple regression and lag model combined with multi-strategy sensitivity analysis method is applied to establish the peak load prediction model in very complex situation. On the basis of modified k-means cluster method,the meteorology characteristic day is grasped,and then preliminary prediction curve is got. The paper judges the deformation probability intelligently and takes optimized correction necessarily; uses daily dynamic data flow to update the modelling parameters to forecast precisely. The proposed method is verified through an application to annual load data of the southern China area, the high accuracy proves its practicability, especially fitting for the complicated and variable weather condition in short time.
作者
李滨
覃芳璐
吴茵
黄佳
Li Bin Qin Fanglu Wu Yin Huang Jia(Guangxi Key Laboratory of Power System Optimization and Energy Technology Guangxi University Nanning 530004 China Power Dispatch Control Center of Guangxi Power Grid Nanning 530023 China)
出处
《电工技术学报》
EI
CSCD
北大核心
2017年第9期149-159,共11页
Transactions of China Electrotechnical Society
基金
国家自然科学基金(51541707)
国家重点基础研究发展计划(973计划)(2013CB228205)资助项目
关键词
短期日负荷预测
完全气象因子
信息粒化
空间多元回归灵敏度
改进K-means聚类
Short-term load forecasting（STLF）
global meteorology factor
information granulated
spatial multiple regression and sensitivity analysis
modified K-means cluster
作者简介
李滨 女,1975年生,博士,副教授,研究方向为电力系统最优化等.E-mail：lizhen@gxu.edu.cn
覃芳璐 女,1991年生,硕士研究生,研究方向为电力系统分析与计算.E-mail：fangluqin@mail.gxu.cn（通信作者）
